pytorch/xla

Pull Requests Commits

[Core Aten Ops] Add test_aten_index_put back (#6370)

ManfeiBai committed 2 years ago

Verified 2a5ad315

Update spmd.md with SPMD debug tool (#6358)

ManfeiBai committed 2 years ago

Verified bc2ebed8

fix core aten asinh (#6365)

zpcore committed 2 years ago

Verified 07832b07

Revert "Fix coverage CI run bug & re-enable the EmbeddingBag test with 1% tol" (#6368)

yeounoh committed 2 years ago

Verified c4b5ab6e

remove random number op aten_randn in the test (#6364)

zpcore committed 2 years ago

Verified 566f3d08

Fix coverage CI run bug & re-enable the EmbeddingBag test with 1% tol (#6361)

yeounoh committed 2 years ago

Verified aa42fee0

Fix some more core aten ops (#6342)

wonjoo-wj committed 2 years ago

Verified 99a1341b

[torchbench] Force `llama_v2_7b_16h` to use SGD. (#6350)

ysiraichi committed 2 years ago

Verified 077b10c1

Improve build process for PJRT plugins (#6314)

will-cromar committed 2 years ago

Verified 61bd1d23

[SPMD] update test format (#6355)

ManfeiBai committed 2 years ago

Verified 269ebcf8

fix issue that torchbench uses xla as device instead of tpu (#6341)

zpcore committed 2 years ago

Verified 0576fe95

Export no-longer accepts callable; wrap in nn.Module. (#6345)

qihqi committed 2 years ago

Verified 1c22fe3a

disable test_EmbeddingBag_per_sample_weights_and_new_offsets (#6351)

JackCaoG committed 2 years ago

Verified e265fd7b

Update README.md for cicleci (#6354)

wonjoo-wj committed 2 years ago

Verified 2e7a84b8

Use SGD for list of models from PyTorch. (#6324)

ysiraichi committed 2 years ago

Verified a7918a72

Allow xla bloat16 torch test results with some variances (#6338)

yeounoh committed 2 years ago

Verified ab28b23f

Update test_spmd_debugging.py to avoid code test code self (#6263)

ManfeiBai committed 2 years ago

Verified df097d75

Update pytorch_test_base.py (#6337)

wonjoo-wj committed 2 years ago

Verified ec23c917

Clean-up CUDA cache. (#6325)

ysiraichi committed 2 years ago

Verified 423bb0b2

Raise an error in `xmp.spawn` if runtime is initialized (#6291)

will-cromar committed 2 years ago

Verified e93597de

Update metric check for resnet18 dynamo unit test (#6307)

wonjoo-wj committed 2 years ago

Verified 8c5e2897

Add option to register installed plugins automatically (#6328)

will-cromar committed 2 years ago

Verified d7ddecdb

Revert "Set all_reduce_token to None when exiting" (#6321)

vanbasten23 committed 2 years ago

Verified f33634e4

Increase absolute tolerance for f16 aten_addmm test (#6332)

LukeBoyer committed 2 years ago

Verified 151dc1d8

Update contribution guide to use python setup.py develop. (#6327)

changm committed 2 years ago

Verified 5f6d2b36

Propagates requires_grad over to AllReduce output (#6326)

yeounoh committed 2 years ago

Verified 40727e4d

Update doc on new modules. (#6318)

golechwierowicz committed 2 years ago

Verified 68515da9

OpenXLA pin update (#6313)

yeounoh committed 2 years ago

Verified 0bd8b8d4

Use spawn as the fork method for the profiler test. (#6302)

ysiraichi committed 2 years ago

Verified 52ef8efc

Add doc images for auto-sharding (#6323)

yeounoh committed 2 years ago

Verified e632897b

Newer Older